Hi: I've got a question about how seed influence the Modeller results? If I start from the same seed in the same computer, does it means it should get the same results? If two job both start from random seed or different seeds, they will get different results, and those results is comparable with running model building jobs with doubled model number building?
THX
On 2/18/11 9:30 AM, albert wrote: > I've got a question about how seed influence the Modeller results? > If I start from the same seed in the same computer, does it means it > should get the same results?
Yes, Modeller is deterministic, so you will get exactly the same results if you run it again in the same environment (same machine type, same operating system).
> If two job both start from random seed
It's not a good idea to start from a random seed, because then your jobs will not be reproducable.
> different seeds, they will get different results, and those results is > comparable with running model building jobs with doubled model number > building?
Yes.
Ben Webb, Modeller Caretaker
participants (2)
-
albert -
Modeller Caretaker